Object motion based video key-frame extraction

We propose a novel method to determine the key-frame of a video, which is based on object motions. We extract foreground objects from a video using optical flow and compute the motion frequency score to find the most representative motion. The key-frame is chosen as the frame that contains the most frequent motion.

[1]  Ying Li,et al.  An Overview of Video Abstraction Techniques , 2001 .

[2]  Simon Baker,et al.  Lucas-Kanade 20 Years On: A Unifying Framework , 2004, International Journal of Computer Vision.

[3]  Vladimir Kolmogorov,et al.  "GrabCut": interactive foreground extraction using iterated graph cuts , 2004, ACM Trans. Graph..

[4]  Andrew Blake,et al.  "GrabCut" , 2004, ACM Trans. Graph..